ffdshow tryouts project: Discussion & Development

[Jeremy Duncan]

Here (http://www.mediafire.com/?zgmfj5gtgny) is a generic patched build using that patch (http://pavelsx.googlepages.com/ffds_avs_patch.diff) (use at your own risk), although I don't even know what that patch does, just to save some time from clsid.

Thank you kindly. :)

I found a trojan in it though: Trojan/Win32.Banbra.gen
This kind of trojan steals banking info for people in Brazil.

Maybe your pc is infected for this trojan to have been in the ffdshow you made for me? :)
So I will not use or tell other people to use this build you kindly made for me for fear it does have a trojan and this is not a false positive.
May I ask clsid to please make me a patch generic build as requested ( link ) (http://forum.doom9.org/showthread.php?p=1336715#post1336715) as I have good experience with his ffdshow builds.

[XhmikosR]

Don't be silly;) It's a known issue with useless AV programs like AVG. Update to the latest signatures, they are supposed to have fixed it. Every Inno Setup installer was detected as a trojan. And as a note, use virustotal.com to check if your AV detects something wrongly or not.
But anyway, too much time spent with you. Maybe I'll infect you too. ;)

EDIT: link removed since you asked for that specific build and you downloaded it.

[Jeremy Duncan]

Don't be silly;) It's a known issue with useless AV programs like AVG. Update to the latest signatures, they are supposed to have fixed it. Every Inno Setup installer was detected as a trojan. And as a note, use virustotal.com to check if your AV detects something wrongly or not.
But anyway, too much time spent with you. Maybe I'll infect you too. ;)

EDIT: link removed since you asked for that specific build and you downloaded it.

I read your post before your edit and tried the svn version and like you said it was detected as having a trojan too.
So either both have trojans or neither do. :p

I'm sorry for offending you. It must be a false positive like you said.

[SamuriHL]

Do you have ANY idea how many false positives AVG produces? It's ridiculous. They must get 100 reports a day about that. It would drive me insane to use something that flags so much stuff as a virus/trojan/whatever. Some people would argue that being overly cautious makes you safer, but, I'd argue against that concept because really, if you can't trust the AV to get it right, then, um, it's not very useful IMO. "Oh, must be another false positive." BAM. :D

[Jeremy Duncan]

Here (http://www.mediafire.com/?zgmfj5gtgny) is a generic patched build using that patch (http://pavelsx.googlepages.com/ffds_avs_patch.diff) (use at your own risk), although I don't even know what that patch does, just to save some time from clsid.

The patched build doesn't work as good as when I use the ffdshow.ax SEt made. I wonder to myself if you applied the patch to ffdshow correctly?
Maybe you can make a new ffdshow.ax with the patch, same build as the one you made for me today and I will see if that works better than the exe.

I'm testing the patched build with mvtools2 motion interpolation, scene skipping in mpc.

[XhmikosR]

No, the process is simple. Patch applied and then rebuilt ffdshow. Setup has the same ffdshow.ax so what you suggest does not make any sense. Nothing more I can do. So ask SEt, or wait until someone takes a look at that patch.

[rica]

jeremy asked.
Pls put his shopping basket what he wants.

duncan, do not forget somebody is trying to help you :cool:

[ryrynz]

@XhmikosR,

Do you plan on continuing to build the core2 versions?

[Sebastiii]

Hi i have receive my 5770 :)
Ready to go to test :)

Seb.

[XhmikosR]

@XhmikosR,

Do you plan on continuing to build the core2 versions?

Maybe, I'll update that build every now and then, but keep in mind that if something is broken, you should use a normal build to see if that one is broken too.
As usual you can find the link in my signature.
If someone could make some more benchmarks, that would be nice, since in my case the performance improvement is small, but not that bad. (see a few pages before)
Also I wonder maybe it's time to update the project files for ICL 11 since no one can download the ICL 10 from Intel. ICL 11.1.048 works very nice, no problems reported. The only thing is that I have to disable vectorization.

[leeperry]

Here is a generic patched build using that patch (http://pavelsx.googlepages.com/ffds_avs_patch.diff) (use at your own risk), although I don't even know what that patch does, just to save some time from clsid.
it breaks the avisynth timecodes to be compatible w/ a heavily broken 2.58 MT DLL and kills 2.57 support(crashes/dropped frames/half speed)...I hope this patch won't make it to the SVN, I don't care for broken 2.58 MT DLL's, 2.57 works pretty fine...and Leak didn't make a patch, meaning it's not required.

plugins need to be compiled w/ the new 2.58 DLL's, not too handy..so let's leave at that until 2.60 comes out.

[XhmikosR]

@leepery: don't tell that to me; tell Jeremy Duncan. I just built a build with that patch, since no one else would do it.

[leeperry]

..for a good reason. hehe, Jeremy likes to play god w/ compiling this DLL, we all have our funs in life ;)

[SEB2010]

Hi guys,

i have a problem with my aspect ratios and i miss an option for the resize/aspect-ratio filter:
I am running WinXP with 800x600 resolution and my drivers and some applications require this resoltution as minimum. Although, my 7" LCD has a native resolution of 800x480 which squeezes the 800x600 diplay-content verticaly. Now i tried to get the right settings in the ffdshow resize-section to get the original aspect ratio on my lcd. This works perfectly for a given aspect-ratio, but is impossible if i e.g. want to watch both 16:9 and 4:3 videos. Only for one aspect i can set the right resize-options.
What i need, is a field, where i can set mulitpliers for one side of the aspect-ratio. For example:
The video is 16:9 and has to be modified to 16:9/480*600=16:11.25 to be displayed right on the LCD. If i play a 4:3 video, the aspect ratio has to be 4:3/480*600=4:3.75 to displayed right.
I thought that the "process pixel aspect ratio internally" might do the trick, but this option had no effect what so ever on my machine.
Is there a combination of configurations which i have to set to get the effect or is this option missing?

thx SEB

[jmartinr]

Hi guys,

i have a problem with my aspect ratios and i miss an option for the resize/aspect-ratio filter:
I am running WinXP with 800x600 resolution and my drivers and some applications require this resoltution as minimum. Although, my 7" LCD has a native resolution of 800x480 which squeezes the 800x600 diplay-content verticaly. Now i tried to get the right settings in the ffdshow resize-section to get the original aspect ratio on my lcd. This works perfectly for a given aspect-ratio, but is impossible if i e.g. want to watch both 16:9 and 4:3 videos. Only for one aspect i can set the right resize-options.
What i need, is a field, where i can set mulitpliers for one side of the aspect-ratio. For example:
The video is 16:9 and has to be modified to 16:9/480*600=16:11.25 to be displayed right on the LCD. If i play a 4:3 video, the aspect ratio has to be 4:3/480*600=4:3.75 to displayed right.
I thought that the "process pixel aspect ratio internally" might do the trick, but this option had no effect what so ever on my machine.
Is there a combination of configurations which i have to set to get the effect or is this option missing?

thx SEB
Set "process pixel aspect ratio internally" and "PAR of output device" to 4:5.

[SEB2010]

thx, played around with this option but never made any difference. only in combination with "keep original aspect ratio" it modified it to the seemingly corret aspect. The resize to fit resolution put the last piece into the puzzle.

thx alot :thanks:

[swordsman]

Whats up with DrawOSD function?:devil:
Will it be back again?

And else - there is no updated delphi header and there is an error (imho) in last IffDecoder.h - the DrawOSD function is removed and added one more shortOSDmessage. When compiling that as overloaded functions - it cause an errors, if I remove old shortOSDmessage function - it works, but without HTML and multiple line handling:confused:

PS I compile this on Delphi 7
Versoin of ffdshow - 3109 & Up

[tal.aloni]

swordsman,
we discussed about it few pages ago,
DrawOSD was required the usage of a registry key to store the OSD message, which made it limited (no more than one message per frame), and hard to maintain with the upgrade to preset based OSD.
we have a new overloaded shortOSDmessage (with x and y coordinates) which is more powerful.

- I wasn't aware of a delphi header, but I see it now (IffDecoder_com.pas), can you use the "Overload;" keyword to make both work in delphi?

otherwise I'm assuming we will have to rename the second shortOSDmessage to something like shortOSDmessageAbsolute.

- regarding HTML / multiple lines, I now see that I've missed something, I'll see about it.

p.s. regarding multiple lines, in most cases you can just use multiple calls to shortOSDmessage.

Thanks,
Tal

[swordsman]

If I use overload function I get errors, if I delete old ShortOSDMessage then I got it working? except HTML & multiline

In ffdshow with drawosd I used next:

.....
st:='<font color=yellow> Sample rate:</font>'+smplrate+'KHz'+#10+
'<font color=yellow> Bit rate:</font>'+bitrate+'KBps'
DrawOSD(10,10, pchar(st));

and got 2 lines with 2 colors.

[tal.aloni]

swordsman,
does + '\n' + between lines works?

p.s.
I'm about to post an updated beta build, I would like your help testing it.

[swordsman]

While inserting \n in user string in ffdshow config it works, but
in remote api it doesn't

[clsid]

Since it is html, maybe <br> will work.

[tal.aloni]

clsid, Great idea, however, it is not supported, and cannot be implemented easily, since html parsing is handled after the lines have been seperated.

[tal.aloni]

swordsman, I was able to fix everything except the multi-line issue, which is a bit complicated. please see if all other problems are solved.

patch:
http://iknowu.net/files/public/ffdshow/PresetOSD/3111-OSD-Bugfixes-beta4.patch

build:
http://iknowu.net/files/public/ffdshow/PresetOSD/ffdshow_rev3111_20091023-OSD-Bugfixes-Beta3.exe

+ bugfix: OSD message format was trimmed after 40 characters.
+ bugfix: HTML tags in OSD messages sent using the OSD API were not applied.
+ enhancement: HTML tags will now be applied on user format as well.
+ internal: the overloaded shortOSDmessage was renamed to shortOSDmessageAbsolute to keep com interface simple.
+ internal: Delphi RemoteAPI sample has been updated.

Thanks,
Tal

Edit: commited in rev. 3113.
Edit2: solution to the multi-line issue was commited in rev 3114. (here is a build for testing purposes: http://iknowu.net/files/public/ffdshow/PresetOSD/ffdshow_rev3114_20091023.exe)

[avivahl]

Tal: your improvements are much appreciated. Toda. :)

[leoliver]

Hi Everyone,

I have ffdshow tryouts revision 2527 on my computer. If I were going to update to the newest version , which I believe is revision 3109 , is it needed to uninstall version 2527 first?
I was hoping to install the new version on top of the older version so I wouldn't have to re-do all of my settings.

[tal.aloni]

leoliver,
you can just upgrade, you don't have to reinstall.

[albain]

Hi all,

my first tool is ready : I will need someone with a xonar to test it


Please PM

Thanks

Damien

[Sebastiii]

Hi albain,

I have just sent PM :)

[albain]

Hi albain,

I have just sent PM :)

OK, I have sent you the link

I hope that you don't have a 64 bits os 'cause I am not sure that it will work (but I guess it will)

[Sebastiii]

Thx :) i've tested but i don't think it works :( because there are not big info in log.
I'm on 32Bits system so it's good :)
Seb.

[madshi]

Thx :) i've tested but i don't think it works :( because there are not big info in log.
I'm on 32Bits system so it's good :)
Seb.
It would help if you actually posted the log. Size isn't everything. Maybe there is something useful in there?

[SamuriHL]

I won't get to test it until much later today if not tomorrow. But I will get to it.

[Sebastiii]

It would help if you actually posted the log. Size isn't everything. Maybe there is something useful in there?

Hi :)

I have put log in PM, but this is the link : Log (http://sebinternet.free.fr/logseb.zip)

I hope that is good :)
Seb.

[albain]

Well, as I answered in the PM, there is no trace at all.

Could you check with a tool called "Process Explorer" from microsoft/sysinternals

After launching the tool and click on launch, launch processxp :
1/ Select view > show lower pane
2/ Select view > Lower pane view > DLLs
3/ Then go to the process of the *well known player that you use* and look for DumpPCM.dll

This is weird that you don't have any trace at all (even the one that are raised after loading)

Otherwise, I have tested it on a windows 7 x64 machine and it works. I just don't have any xonar either on that one :):)

*EDIT* : maybe you should disable your firewall during all this.

If the load is done, you should be able to see :
Got graph

and then when playing the media :
{..............some GUID...........} "<a name> / IN : # / OUT : #

and then the dump traces

[madshi]

@albain, you could add some more logging to the dll. I'd suggest to log whether "HookAPI(CoCreateInstanceEx)" returns true or false. Also you could add some logging to all of the hook callback functions. E.g. you could log all of the GUIDs that CoCreateInstanceEx is called for (see "SysUtils.GuidToString(...)"). It's possible that the media player isn't using CoCreateInstanceEx to instantiate the filter graph. Maybe it's using some other way. Maybe it's calling CoCreateInstanceEx for something else and then uses QueryInterface to get the filter graph? Don't know...

[Sebastiii]

I'll try and idem, with process explorer i can see the DumpPCM.dll and the link is ok :

D:\Users\Homecinema\Desktop\dumper_tool\DumpPCM.dll

But in the log, that is strange i think it don't dump the real sound card :
[2536] CRealtekCP::CRealtekCP()

But i'm not a programmer lol so i must be wrong.

Even if i disable in bios the sound card realtek maybe something keep in registry.

Seb.

[albain]

Okay, maybe the player is using another way to instantiate the graph for bitstream mode but this is still weird because just the start of the player generates "Got graph" traces

I will add some more traces though

[SamuriHL]

I'm installing the driver now so I'll be able to test in a few minutes on my machine.

[SamuriHL]

http://pastebin.com/pastebin.php?dl=d47c292b1

The TrueHD sample you included. Does this help? If yes, I'll jam on a DTS-HD MA file next.

[albain]

Sorry, this is not what I expect. I'll come back to you with a fix

[Sebastiii]

Good luck :)

[SamuriHL]

albain, I'm sure you're aware of this but you are up against thermida, PAP, and other protections they use for the audio path. :)

[leoliver]

leoliver,
you can just upgrade, you don't have to reinstall.

Thanks for your reply.

[whurlston]

albain, are you just trying to get the media types used? I have successfully exposed the graph used by ArcSoft TMT by installing FilterGraph Spy: http://alax.info/blog/777

I've also connected to the graph using GraphStudio but I still can't get the Xonar to work on my setup so I can't help with that part.

[albain]

albain, are you just trying to get the media types used? I have successfully exposed the graph used by ArcSoft TMT by installing FilterGraph Spy: http://alax.info/blog/777

I've also connected to the graph using GraphStudio but I still can't get the Xonar to work on my setup so I can't help with that part.

Not just that, also the content of the buffers that are exchanged between the decoder and the renderer.
But your link may help as an alternative but I expect counter measures against that

For now I am working on improving the current hook tool with madshi's help

[SamuriHL]

Not just that, also the content of the buffers that are exchanged between the decoder and the renderer.
But your link may help as an alternative but I expect counter measures against that

For now I am working on improving the current hook tool with madshi's help

Excellent. When you have something new, let me know. I've left the Xonar installed and set up for this very purpose.

[Cheesegeezer]

Do you need a tester for the HD58xx series cards. I'm here and available

[Sebastiii]

Excellent. When you have something new, let me know. I've left the Xonar installed and set up for this very purpose.

Yes excellent, i'm here too with my Xonar :)

[albain]

Hi,

we have difficulties to make the hook tool work with xonar owners, maybe due to a protection that is activated for this case, this would explain why the tool works for me

Until we figure this out, here is a new FFDShow beta build (http://damienbt.free.fr/ffdshow_rev3115_20091025_dbt_bitstream_beta26.zip) for HD bitstream.

Thanks for your patience